Skip to content

Latest commit

 

History

History
34 lines (26 loc) · 1.78 KB

Texture_Synthesis_Through_CNNs_and_Spectrum_Constraints.md

File metadata and controls

34 lines (26 loc) · 1.78 KB

Paper

  • Title: Texture Synthesis Through Convolutional Neural Networks and Spectrum Constraints
  • Authors: Gang Liu, Yann Gousseau, Gui-Song Xia
  • Link: http://arxiv.org/abs/1605.01141v3
  • Tags: Neural Network, artistic style, fourier, texture
  • Year: 2016

Summary

  • What

    • The well known method of Artistic Style Transfer can be used to generate new texture images (from an existing example) by skipping the content loss and only using the style loss.
    • The method however can have problems with large scale structures and quasi-periodic patterns.
    • They add a new loss based on the spectrum of the images (synthesized image and style image), which decreases these problems and handles especially periodic patterns well.
  • How

    • Everything is handled in the same way as in the Artistic Style Transfer paper (without content loss).
    • On top of that they add their spectrum loss:
      • The loss is based on a squared distance, i.e. 1/2 d(I_s, I_t)^2.
        • I_s is the last synthesized image.
        • I_t is the texture example.
      • d(I_s, I_t) then does the following:
        • It assumes that I_t is an example for a space of target images.
        • Within that set it finds the image I_p which is most similar to I_s. That is done using a projection via Fourier Transformations. (See formula 5 in the paper.)
        • The returned distance is then I_s - I_p.
  • Results

    • Equal quality for textures without quasi-periodic structures.
    • Significantly better quality for textures with quasi-periodic structures.

Overview

Overview over their method, i.e. generated textures using style and/or spectrum-based loss.